Final Test Flashcards
(28 cards)
What does elaboration mean?
Looking at what happens to a bivariate (zero-order) relationships when you control for a third variable - does it change, by how much and for whom, or does it stay the same?
ex. could add gender as a 3rd variable > put all the women “away” and just look at the men and see what happens
Explain how elaboration impacted the Caffeine and Miscarriage study
Bivariate Results:
- looking just at caffeine intake and miscarriages, the tentative conclusion was that more than 200mg of caffeine almost doubles a woman’s chance of miscarriage
Control Variable (added control variable of nauseated / non-nauseated) - seeing if this variable changes the caffeine to miscarriage relationship
1st Partial Results (non-nauseated group):
- maximum differences are basically the same as zero-order (both moderate relationships)
2nd Partial Results (nauseated group):
- maximum differences still the same as zero-order (also moderate)
**Results = the zero-order relationship still holds > reinforces the original relationship
The model is REPLICATION
What is a zero-order relationship?
It is the bivariate relationships where an independent variable is the proposed cause and the dependent variable is the proposed effect
What are the 4 different elaboration models?
Replication: Partials & original order are the same (same max difference strength)
Explanation / Spuriousness: Partial relationships are 0 or less than original order & the CV is ANTECEDENT
Interpretation: Partial relationships are 0 or less than original order & CV is INTERVENING
Specification: One partial relationship is the same as or greater than the original order & other partial is 0 or less than the original order
What are the 5 steps to determining the name of your elaboration model?
- Summarize the zero-order relationship
- Summarize each of your partial relationships
- Compare the partials with the zero-order results (compare the names - weak, moderate, or strong)
- Determine the Time Order
- Name your model (replication, explanation, interpretation, specification)
What does an antecedent and intervening control variable mean?
Antecedent: the control variable is acting before both the independent and dependent variables (CV > IV/DV)
Intervening: the control variable is acting after the independent variable but before the dependent variable (IV > CV > DV)
Describe the scatterplot association between female literacy rate and fertility rate - so what kind of statistic can we use for this association?
An increase in the female literacy rate is associated with a decrease in the fertility rate (higher literacy delays child bearing and reduces fertility rate)
- scatterplot looks like a negative relationship in the beggining but eventually there are diminishing returns > it’s not a fully linear relationship because we can’t say it’s completely negative
**we can use Spearman Rank Correlation for this relationship BECUASE it’s not fully linear
What is monotonicity and how is it different from linearity?
As one variable changes, the other tends to change consistently (either consistently increases or consistently decreases, without changing direction)
With linearity, the slope should be constant (straight line) - no curves in the relationship - no diminishing or increasing rates of change
What are the assumptions we make before being able to use Spearman Rank Correlation?
- random sample
- at least ordinal level measurement
- variables that change monotonically with each other (if they’re linear then we can probably use Pearson’s r)
- usually we also have quite a few categories of variables (which is why we don’t just use the cross-tab analysis methods)
How do you find the PRE score with Spearman’s rho?
You take the answer you got from spearman’s rho (r) and you square that number (decreases it a bit)
What are scatterplots - why are they useful?
aka Scattergrams
It is a graph upon which a researcher plots each case or observation, and each axis represents the value of one variable (a dot = the intersection of two variables)
- independent variable on x axis
- dependent variable on y axis
Scatterplots help give us an initial sense of direction and strength of the relationship (dots less spread out and more uniform = stronger relationship)
What are the 3 characteristics we can see from a scatterplot?
- what is the existence and strength of the relationship (how concentrated are the dots around the through centre / best fit line)
- what is the direction
- what is the extent of linearity
What would a perfect relationship look like on a scatterplot graph?
Completely straight and linear line - dots all fall right on the line perfectly
What is a zero relationship?
aka Independence
- an increase/decrease in x doesn’t impact y value at all
What is the relationship between linearity and direction?
Positive/negative relationships are by definition, linear
- linearity is a necessary condition for a direction
If you have a non-linear bivariate scatterplot, you should use _________ for the measure of association
spearman rank correlation
What is linear regression (ordinary least squares regression)? (what does it allow us to do)
It allows us to describe relationships more precisely - makes a predictive equation (predict y’ from a given x)
- constructs a model that best fits the data (regression calculates the best fit line)
- finds the strength and direction of a relationship
- tend to use in conjunction with Pearson’s r which uses similar statistical techniques and assumptions
What are the assumptions needed for OLS regression and Pearson’s r (6)
Assumes that your data are/have…
- come from a random sample
- continuous interval/ratio level of measurement
- normal distributions
- absence of significant outliers
- linear bivariate association
absence of non-sampling errors
What is the regression line equation?
Y = a + bX
Y = score on the dependent variable
a = the Y intercept
b = the slope of the regression line
X = score on the independent variable
What does Y’ indicate?
It indicates a predicted value
What does multiple regression allow?
It allows us to control for additional variables
How is Regression similar and different from Pearson’s r?
They both describe the direction and strength of a relationship
Regression can infer (predict) specific values of Y when we have values of X but Pearson’s r cannot do this (descriptive AND inferential)
Linear regression has both ________ and _______ aspects
descriptive
inferential
What does Pearson’s r allow us to measure? What does it require?
Allows us to measure the direction and strength of association between 2 interval/ratio variables
- it is symmetrical - doesn’t really explain cause and effect
Use PRE logic > whereas Lambda guessed Y using the mode, Pearson’s r guessed Y using the mean
Generally requires the same assumptions to be met that regression requires
